coffee/libyuv
1
0
Fork 0
mirror of https://chromium.googlesource.com/libyuv/libyuv synced 2025-12-08 01:36:47 +08:00
Code Issues Packages Projects Releases Wiki Activity

ARGBToI444_SSSE3 port to gcc

Browse Source 
BUG=148
TEST=try bots
Review URL: https://webrtc-codereview.appspot.com/1091004

git-svn-id: http://libyuv.googlecode.com/svn/trunk@557 16f28f9a-4ce2-e073-06de-1de4eb20be90
This commit is contained in:
fbarchard@google.com 2013-02-04 18:47:21 +00:00
parent 92352b7081
commit 762c050b65
5 changed files with 125 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
README.chromium
View File

@ -1,6 +1,6 @@
Name: libyuv
URL: http://code.google.com/p/libyuv/
Version: 556
Version: 557
License: BSD
License File: LICENSE

2
include/libyuv/row.h
View File

@ -49,6 +49,7 @@ extern "C" {
#define HAS_ARGBTORGB565ROW_SSE2
#define HAS_ARGBTORGBAROW_SSSE3
#define HAS_ARGBTOUV422ROW_SSSE3
#define HAS_ARGBTOUV444ROW_SSSE3
#define HAS_ARGBTOUVROW_SSSE3
#define HAS_ARGBTOYROW_SSSE3
#define HAS_BGRATOARGBROW_SSSE3
@ -121,7 +122,6 @@ extern "C" {
// TODO(fbarchard): Port to gcc.
#if !defined(YUV_DISABLE_ASM) && defined(_M_IX86)
#define HAS_ARGBCOLORTABLEROW_X86
#define HAS_ARGBTOUV444ROW_SSSE3
#define HAS_ARGBINTERPOLATEROW_SSE2
#endif

2
include/libyuv/version.h
View File

@ -11,6 +11,6 @@
#ifndef INCLUDE_LIBYUV_VERSION_H_ // NOLINT
#define INCLUDE_LIBYUV_VERSION_H_
#define LIBYUV_VERSION 556
#define LIBYUV_VERSION 557
#endif // INCLUDE_LIBYUV_VERSION_H_ NOLINT

122
source/row_posix.cc
View File

@ -925,6 +925,128 @@ void ARGBToUVRow_Unaligned_SSSE3(const uint8* src_argb0, int src_stride_argb,
);
}
void ARGBToUV444Row_SSSE3(const uint8* src_argb, uint8* dst_u, uint8* dst_v,
int width) {
asm volatile (
"movdqa %0,%%xmm4 \n"
"movdqa %1,%%xmm3 \n"
"movdqa %2,%%xmm5 \n"
:
: "m"(kARGBToU), // %0
"m"(kARGBToV), // %1
"m"(kAddUV128) // %2
);
asm volatile (
"sub %1,%2 \n"
".p2align 4 \n"
"1: \n"
"movdqa (%0),%%xmm0 \n"
"movdqa 0x10(%0),%%xmm1 \n"
"movdqa 0x20(%0),%%xmm2 \n"
"movdqa 0x30(%0),%%xmm6 \n"
"pmaddubsw %%xmm4,%%xmm0 \n"
"pmaddubsw %%xmm4,%%xmm1 \n"
"pmaddubsw %%xmm4,%%xmm2 \n"
"pmaddubsw %%xmm4,%%xmm6 \n"
"phaddw %%xmm1,%%xmm0 \n"
"phaddw %%xmm6,%%xmm2 \n"
"psrlw $0x8,%%xmm0 \n"
"psrlw $0x8,%%xmm2 \n"
"packuswb %%xmm2,%%xmm0 \n"
"paddb %%xmm5,%%xmm0 \n"
"sub $0x10,%3 \n"
"movdqa %%xmm0,(%1) \n"
"movdqa (%0),%%xmm0 \n"
"movdqa 0x10(%0),%%xmm1 \n"
"movdqa 0x20(%0),%%xmm2 \n"
"movdqa 0x30(%0),%%xmm6 \n"
"pmaddubsw %%xmm3,%%xmm0 \n"
"pmaddubsw %%xmm3,%%xmm1 \n"
"pmaddubsw %%xmm3,%%xmm2 \n"
"pmaddubsw %%xmm3,%%xmm6 \n"
"phaddw %%xmm1,%%xmm0 \n"
"phaddw %%xmm6,%%xmm2 \n"
"psrlw $0x8,%%xmm0 \n"
"psrlw $0x8,%%xmm2 \n"
"packuswb %%xmm2,%%xmm0 \n"
"paddb %%xmm5,%%xmm0 \n"
"lea 0x40(%0),%0 \n"
"movdqa %%xmm0,(%1,%2,1) \n"
"lea 0x10(%1),%1 \n"
"jg 1b \n"
: "+r"(src_argb), // %0
"+r"(dst_u), // %1
"+r"(dst_v), // %2
"+rm"(width) // %3
:
: "memory", "cc"
#if defined(__SSE2__)
, "xmm0", "xmm1", "xmm2", "xmm6"
#endif
);
}
void ARGBToUV444Row_Unaligned_SSSE3(const uint8* src_argb, uint8* dst_u,
uint8* dst_v, int width) {
asm volatile (
"movdqa %0,%%xmm4 \n"
"movdqa %1,%%xmm3 \n"
"movdqa %2,%%xmm5 \n"
:
: "m"(kARGBToU), // %0
"m"(kARGBToV), // %1
"m"(kAddUV128) // %2
);
asm volatile (
"sub %1,%2 \n"
".p2align 4 \n"
"1: \n"
"movdqu (%0),%%xmm0 \n"
"movdqu 0x10(%0),%%xmm1 \n"
"movdqu 0x20(%0),%%xmm2 \n"
"movdqu 0x30(%0),%%xmm6 \n"
"pmaddubsw %%xmm4,%%xmm0 \n"
"pmaddubsw %%xmm4,%%xmm1 \n"
"pmaddubsw %%xmm4,%%xmm2 \n"
"pmaddubsw %%xmm4,%%xmm6 \n"
"phaddw %%xmm1,%%xmm0 \n"
"phaddw %%xmm6,%%xmm2 \n"
"psrlw $0x8,%%xmm0 \n"
"psrlw $0x8,%%xmm2 \n"
"packuswb %%xmm2,%%xmm0 \n"
"paddb %%xmm5,%%xmm0 \n"
"sub $0x10,%3 \n"
"movdqu %%xmm0,(%1) \n"
"movdqu (%0),%%xmm0 \n"
"movdqu 0x10(%0),%%xmm1 \n"
"movdqu 0x20(%0),%%xmm2 \n"
"movdqu 0x30(%0),%%xmm6 \n"
"pmaddubsw %%xmm3,%%xmm0 \n"
"pmaddubsw %%xmm3,%%xmm1 \n"
"pmaddubsw %%xmm3,%%xmm2 \n"
"pmaddubsw %%xmm3,%%xmm6 \n"
"phaddw %%xmm1,%%xmm0 \n"
"phaddw %%xmm6,%%xmm2 \n"
"psrlw $0x8,%%xmm0 \n"
"psrlw $0x8,%%xmm2 \n"
"packuswb %%xmm2,%%xmm0 \n"
"paddb %%xmm5,%%xmm0 \n"
"lea 0x40(%0),%0 \n"
"movdqu %%xmm0,(%1,%2,1) \n"
"lea 0x10(%1),%1 \n"
"jg 1b \n"
: "+r"(src_argb), // %0
"+r"(dst_u), // %1
"+r"(dst_v), // %2
"+rm"(width) // %3
:
: "memory", "cc"
#if defined(__SSE2__)
, "xmm0", "xmm1", "xmm2", "xmm6"
#endif
);
}
void ARGBToUV422Row_SSSE3(const uint8* src_argb0,
uint8* dst_u, uint8* dst_v, int width) {
asm volatile (

3
source/scale_argb.cc
View File

@ -23,9 +23,6 @@ namespace libyuv {
extern "C" {
#endif
// Bilinear SSE2 is disabled.
#define SSE2_DISABLED 1
// ARGB scaling uses bilinear or point, but not box filter.
#if !defined(YUV_DISABLE_ASM) && (defined(__ARM_NEON__) || defined(LIBYUV_NEON))